Cut out and Replay: A Simple yet Versatile Strategy for Multi-Label Online Continual Learning

Xinrui Wang, Shao-Yuan Li, Jiaqiang Zhang, Songcan Chen
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:63530-63548, 2025.

Abstract

Multi-Label Online Continual Learning (MOCL) requires models to learn continuously from endless multi-label data streams, facing complex challenges including persistent catastrophic forgetting, potential missing labels, and uncontrollable imbalanced class distributions. While existing MOCL methods attempt to address these challenges through various techniques, they all overlook label-specific region identifying and feature learning - a fundamental solution rooted in multi-label learning but challenging to achieve in the online setting with incremental and partial supervision. To this end, we first leverage the inherent structural information of input data to evaluate and verify the innate localization capability of different pre-trained models. Then, we propose CUTER (CUT-out-and-Experience-Replay), a simple yet versatile strategy that provides fine-grained supervision signals by further identifying, strengthening and cutting out label-specific regions for efficient experience replay. It not only enables models to simultaneously address catastrophic forgetting, missing labels, and class imbalance challenges, but also serves as an orthogonal solution that seamlessly integrates with existing approaches. Extensive experiments on multiple multi-label image benchmarks demonstrate the superiority of our proposed method. The code is available at https://github.com/wxr99/Cut-Replay

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-wang25bg, title = {Cut out and Replay: A Simple yet Versatile Strategy for Multi-Label Online Continual Learning}, author = {Wang, Xinrui and Li, Shao-Yuan and Zhang, Jiaqiang and Chen, Songcan}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {63530--63548}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/wang25bg/wang25bg.pdf}, url = {https://proceedings.mlr.press/v267/wang25bg.html}, abstract = {Multi-Label Online Continual Learning (MOCL) requires models to learn continuously from endless multi-label data streams, facing complex challenges including persistent catastrophic forgetting, potential missing labels, and uncontrollable imbalanced class distributions. While existing MOCL methods attempt to address these challenges through various techniques, they all overlook label-specific region identifying and feature learning - a fundamental solution rooted in multi-label learning but challenging to achieve in the online setting with incremental and partial supervision. To this end, we first leverage the inherent structural information of input data to evaluate and verify the innate localization capability of different pre-trained models. Then, we propose CUTER (CUT-out-and-Experience-Replay), a simple yet versatile strategy that provides fine-grained supervision signals by further identifying, strengthening and cutting out label-specific regions for efficient experience replay. It not only enables models to simultaneously address catastrophic forgetting, missing labels, and class imbalance challenges, but also serves as an orthogonal solution that seamlessly integrates with existing approaches. Extensive experiments on multiple multi-label image benchmarks demonstrate the superiority of our proposed method. The code is available at https://github.com/wxr99/Cut-Replay} }
Endnote
%0 Conference Paper %T Cut out and Replay: A Simple yet Versatile Strategy for Multi-Label Online Continual Learning %A Xinrui Wang %A Shao-Yuan Li %A Jiaqiang Zhang %A Songcan Chen %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-wang25bg %I PMLR %P 63530--63548 %U https://proceedings.mlr.press/v267/wang25bg.html %V 267 %X Multi-Label Online Continual Learning (MOCL) requires models to learn continuously from endless multi-label data streams, facing complex challenges including persistent catastrophic forgetting, potential missing labels, and uncontrollable imbalanced class distributions. While existing MOCL methods attempt to address these challenges through various techniques, they all overlook label-specific region identifying and feature learning - a fundamental solution rooted in multi-label learning but challenging to achieve in the online setting with incremental and partial supervision. To this end, we first leverage the inherent structural information of input data to evaluate and verify the innate localization capability of different pre-trained models. Then, we propose CUTER (CUT-out-and-Experience-Replay), a simple yet versatile strategy that provides fine-grained supervision signals by further identifying, strengthening and cutting out label-specific regions for efficient experience replay. It not only enables models to simultaneously address catastrophic forgetting, missing labels, and class imbalance challenges, but also serves as an orthogonal solution that seamlessly integrates with existing approaches. Extensive experiments on multiple multi-label image benchmarks demonstrate the superiority of our proposed method. The code is available at https://github.com/wxr99/Cut-Replay
APA
Wang, X., Li, S., Zhang, J. & Chen, S.. (2025). Cut out and Replay: A Simple yet Versatile Strategy for Multi-Label Online Continual Learning. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:63530-63548 Available from https://proceedings.mlr.press/v267/wang25bg.html.

Related Material